It is recognized that the performance of any prediction model is a function of several factors. One of the most significant factors is the adopted preprocessing techniques. In other words, preprocessing is an essential process to generate an effective and efficient classification model. This paper investigates the impact of the most widely used preprocessing techniques, with respect to numerical features, on the performance of classification algorithms. The effect of combining various normalization techniques and handling missing values strategies is assessed on eighteen benchmark datasets using two well-known classification algorithms and adopting different performance evaluation metrics and statistical significance tests. According to the reported experimental results, the impact of the adopted preprocessing techniques varies from one classification algorithm to another. In addition, a statistically significant difference between the considered data preprocessing techniques is demonstrated.

Chest and lung diseases are among the most serious chronic diseases in the world, and they occur as a result of factors such as smoking, air pollution, or bacterial infection, which would expose the respiratory system and chest to serious disorders. Chest diseases lead to a natural weakness in the respiratory system, which requires the patient to take care and attention to alleviate this problem. Countries are interested in encouraging medical research and monitoring the spread of communicable diseases. Therefore, they advised researchers to perform studies to curb the diseases’ spread and urged researchers to devise methods for swiftly and readily detecting and distinguishing lung diseases. In this paper, we propose a hybrid architecture of contrast-limited adaptive histogram equalization (CLAHE) and deep convolutional network for the classification of lung diseases. We used X-ray images to create a convolutional neural network (CNN) for early identification and categorization of lung diseases. Initially, the proposed method implemented the support vector machine to classify the images with and without using CLAHE equalizer. The obtained results were compared with the CNN networks. Later, two different experiments were implemented with hybrid architecture of deep CNN networks and CLAHE as a preprocessing for image enhancement. The experimental results indicate that the suggested hybrid architecture outperforms traditional methods by roughly 20% in terms of accuracy.

Teaching and exam proctoring represent key pillars of the education system. Human proctoring, which involves visually monitoring examinees throughout exams, is an important part of assessing the academic process. The capacity to proctor examinations is a critical component of educational scalability. However, such approaches are time-consuming and expensive. In this paper, we present a new framework for the learning and classification of cheating video sequences. This kind of study aids in the early detection of students’ cheating. Furthermore, we introduce a new dataset, “actions of student cheating in paper-based exams”. The dataset consists of suspicious actions in an exam environment. Five classes of cheating were performed by eight different actors. Each pair of subjects conducted five distinct cheating activities. To evaluate the performance of the proposed framework, we conducted experiments on action recognition tasks at the frame level using five types of well-known features. The findings from the experiments on the framework were impressive and substantial.

Knee Osteoarthritis (OA) is considered as one of the most popular diseases for elder people with the age of 60 and upper. In addition, there are over 10 million people in Thailand were affected by Knee OA. Age and weight are the two main risk factors to make people have knee OA. When OA has appeared to the patient, it is totally difficult to recovery back as the normal. Thus, knee OA early detection is the most important factor to prevent knee OA. The typical way to detect and analyze OA is the X-ray imaging application. This research study is directed to the early detection of OA by applying image processing (specifically the shape analysis) applied with classification techniques to knee X-ray imagery. The basic idea of the work is to find a region of interest, use shape decomposition technique and build a classifier that can later classify between OA or non-OA imageries. Firstly, the quadtree decomposition is applied to X-ray images for analyzing the regions of interest (ROI) of the knee X-ray image. There are three data sets include (i) whole knee (Dataset 1), (ii) knee joint space ROI (Dataset 2), and (iii) the application Otsu's method to knee joint space ROI (Dataset 3). Secondly, feature selection is adopted to this work in order to reduce the feature space in terms of the number of values and dimensions. Lastly, the classifier generator is applied to generate the desired classifieds which can be used to classify knee images between non-OA (normal case) and OA. The challenge of the study is how to know which ROI and the threshold value in shape decomposition are suitable to use for the classification process. The data were obtained 128 male and female participants were used for the evaluation while OA imagery was presented as 66 images. The results obtained show that the Dataset 3 sub-image is the most appropriate region of interest to consider come along with the threshold value of 10. The best classification performance in term of an Area Under the ROC Curve (AUC) value of 0.917 was recorded.

Bird’s-eye surround view is a new type of Advanced Driv er Assistance System (ADASs) that provides drivers with a real-time ${360}^{\circ }$ top-down view of their vehicle. This paper presents an architecture for a surround view system based on a Tensilica Vision P5 embedded DSP. Fish-eye cameras mounted on each side of a vehicle are employed in the proposed model. The cameras are able to cover the entire view around the vehicle with overlapping areas between neighboring cameras. We use a morphing approach for joint lens and perspective correction of each view. For stitching frames from different cameras, we assume a fixed geometry which permits off-line calculation of frame transformations, thereby reducing the real-time processing load. To further reduce computational complexity, we employ lookup-tables in the view morphing and blending phases. The Vision P5 core combines the four frames with XGA resolution into a single XGA surround view at 30FPS.